I have embraced the Cookie AutoDelete way, along with container tabs. I log in again, therefore I am :)

I concur: the perennial solution should happen SMBC-side, through careful analysis of assigned cookies (but analyzing past cookies can be difficult). It may even be possible to get rid of worst offenders through a well-crafted Set-Cookie header. Increasing the limit remains a transitional option.

Yes, this matches my technical suspicions too. As to the number and sketchiness of ads, I can only trust you as I discovered their very existence while taking a look at this issue.

TL;DR: clean your cookies
More in this thread: bsky.app/profile/zach...

Is it just me or is the terminology in that article confusing?

Alas, I predict they are going to love the metadata in the published JPEG file -- images.nasa.gov/details/art0...

Do not the DNS!

By disabling adblock, I immediately reached 1 KiB worth of cookies, with `euconsent-v2` being quite long.
But smbc-comics.com currently supports up to 8 KiB headers, so I still wonder what exact cookies piled up and reached that threshold.

@zachweinersmith.bsky.social Everyone's hunch so far: analytics and advertising-related cookies pile up in some browsers (presumably those with no adblock whatsoever) until your server says "sorry, sir, you are bringing way too many cookies for my feeble appetite".

My bad, I must have mixed your report with someone else's in that thread :)

Use your browser's developer tools to inspect cookies (typically found in the "Storage" section); deleting them should solve the issue, BUT analyzing cookies themselves (names, contents, length, domains and tools they relate to) could help everyone.

Use your browser's developer tools to inspect cookies (typically found in the "Storage" section); deleting them should solve the issue, BUT analyzing cookies themselves (names, contents, length, domains and tools they relate to) could help everyone.

Well. Presumably, that was an excess of third-party cookies. I assume all other websites affected by that issue were also .com domains?

This is likely fixed... until the same issue happens again next time. Did you inspect cookies sent to that domain and path before clearing them? As much as HTTP 400 Bad Request is considered a client-side issue, cookies are assigned by the server to the client, i.e. there may be a server-side issue.

> or are actively avoiding those things.
If true, that part is particularly bleak.

I used to think that. Today, I would say it heavily depends on implementation and user habits. Specifically, modern and secure account management relies on our devices remaining available. Steal, burn or fry ALL my hardware at once and I am effectively locked out of virtually all my online accounts.

Actually, that would be absolutely lovable outside the current shitopia. As long as society's response to "my job has just been automated" is "tough luck", automation is bound to spark opposition.

I would suggest a stroll through the gazillion features that modern CSS offers, along with keywords to explore them. That could prevent beginners from reinventing the wheel.

Thought #1: can one even install claude without `curl|bash`? brew comes to mind but it also suggests some bash+curl install command.
Thought #2: considering claude takes human-readable input, they could at least suggest `curl | vipe | claude` (from the moreutils package).

The evil blind bird is hungry too: `> important_document.txt`

Of the tremendous semantic importance of adverbs.

... which suddenly makes the diagram more complex to produce and read, eh? :)

Sooo... what if the decorator requires arguments?

> If there's any single DO NOT DO between AGI and Apocalypse we are toast.
So... does "Do not trigger the apocalypse" count?

Frowning in nnoremap <C-d> :q<CR>

Is this officially called the middle finger distribution?

Uh? We sound worse. Orders of magnitude worse.

Now make it work with an arbitrary integer base :)

I say "I did not take the time to ...". That typically implies I could have dedicated some time to the task without breaking the law of physics but chose not to, also implying my time was spent on other priorities.

Worth mentioning the exit code should be kept below 255? System-wise, only "the least significant 8 bits" are kept.